AIConf
This project provides convenience functions to handle config files in HOCON, in particular for the selection of the use of implemented classes and their parameter orchestration.
Installation
pip install AIconfExample
Say you have multiple implementations of SomeInterface
class SomeInterface:
...
class SomeClass(SomeInterface):
def __init__(self, arg1, arg2):
...
class SomeOtherClass(SomeInterface):
def __init__(self, arg3, arg4):
...You want to use a config file sample.conf to select appropriate implementations transparently and choose appropriate parameters.
impl_1 = {
'()' = somepackage.SomeClass
arg1 = 3
arg2 = 4
}
impl_2 = {
'()': somepackage.SomeOtherClass
arg3 = true
arg4 = false
}
calling
from aiconf import ConfigReader
from aiconf import construct_from_config
cfg = ConfigReader('sample.conf').read_config()
impl_1 = construct_from_config(cfg['impl_1'])
impl_2 = construct_from_config(cfg['impl_2'])
assert isinstance(impl_1, SomeClass)
assert isinstance(impl_1, SomeOtherClass)Config file backed Argument parser
With a config file config.conf
foo = "bar"
d {
y = 3
z = 4
}
and a python script test.py
from aiconf import get_arg_parser
parser = get_arg_parser("This is an example arg parser", 'config.conf')
# works like a usual parser
parser.add_argument("--wonk", type=int, default=0)
# get args as usual
args = parser.parse_args()
# get access to the config
conf = parser.conf
assert args.wonk == 3
assert conf['foo'] == "bar"
assert conf['d.y'] == 5run
python test.py --wonk 3 -c foo=bar -c d.y = 5marvel at its beauty and the merged configs.
For further information, refer to the documentation.
